Training Robots to Evaluate Robots: Example-Based Interactive Reward Functions for Policy Learning
This addresses the challenge of reward specification in robotics by enabling robots to learn evaluation behaviors from examples, potentially reducing manual engineering for tasks like assembly or manipulation.
The paper tackles the problem of training robots to automatically acquire interactive behaviors for evaluating skill execution, which then serve as interactive reward functions (IRFs) to train reinforcement learning policies for tasks like screw tightening. The result shows large performance improvements in evaluations on door locking, weighted block stacking, and screw tightening, outperforming baselines with demonstrations or engineered rewards.
Physical interactions can often help reveal information that is not readily apparent. For example, we may tug at a table leg to evaluate whether it is built well, or turn a water bottle upside down to check that it is watertight. We propose to train robots to acquire such interactive behaviors automatically, for the purpose of evaluating the result of an attempted robotic skill execution. These evaluations in turn serve as "interactive reward functions" (IRFs) for training reinforcement learning policies to perform the target skill, such as screwing the table leg tightly. In addition, even after task policies are fully trained, IRFs can serve as verification mechanisms that improve online task execution. For any given task, our IRFs can be conveniently trained using only examples of successful outcomes, and no further specification is needed to train the task policy thereafter. In our evaluations on door locking and weighted block stacking in simulation, and screw tightening on a real robot, IRFs enable large performance improvements, even outperforming baselines with access to demonstrations or carefully engineered rewards. Project website: https://sites.google.com/view/lirf-corl-2022/